Know Your Safe Drinking Skills: Adaptation Strategies for the College Effect - MDPI
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
$
£ ¥€
social sciences
Article
Know Your Safe Drinking Skills: Adaptation Strategies for the
College Effect
Carolyn A. Lin * , John L. Christensen and Anne Borsai Basaran
Department of Communication, University of Connecticut, Storrs, Mansfield, CT 06269-1295, USA;
john.christensen@uconn.edu (J.L.C.); anne.borsai@uconn.edu (A.B.B.)
* Correspondence: carolyn.lin@uconn.edu
Abstract: Objective: The current study investigates the effects of an alcohol-prevention program
delivered to college students in a formal classroom setting. Participants: The sample comprised
231 first-year college students who enrolled in a multisection “First Year Experience” course at a large
northeastern university in the United States. Method: A naturalistic experiment was conducted, with
a baseline evaluation at the beginning of the semester and a post-experiment evaluation near the end
of the semester. Results: Social drinking attitudes, proximal drinking norm and the college effect are
significant predictors of pre- and post-intervention episodic drinking frequency. The intervention
reduced episodic drinking frequency as well as perceived distal and proximal drinking norms. It also
increased drinking attitudes and did not change perceived efficacy or drinking-outcome expectancies.
Conclusions: Practitioners could consider implementing a similar intervention to allow students to
learn and practice safe drinking skills in the first year of their college life.
Keywords: adaptation strategies; alcohol prevention; college effect; naturalistic experiment; theory
of planned behavior
Citation: Lin, Carolyn A., John L.
Christensen, and Anne Borsai
Basaran. 2022. Know Your Safe
Drinking Skills: Adaptation
1. Introduction
Strategies for the College Effect. Episodic drinking behavior is a significant health problem on American college cam-
Social Sciences 11: 18. https:// puses. According to the National Survey on Drug Use and Health (NSDUH), 53 and 33%
doi.org/10.3390/socsci11010018 of full-time U.S. college students between the ages of 18–22 reported regularly drinking
alcoholic beverages and engaging in binge drinking, respectively (SAMHSA 2019). The U.S.
Academic Editor: Timo
Juhani Lajunen
health agency, National Institute on Alcohol Abuse and Alcoholism, defines binge drinking
as a drinking pattern that raises the blood alcohol concentration of an adult to 0.08 percent
Received: 16 October 2021 (or 0.08 g of alcohol per deciliter) or higher, through consuming five or more drinks (male)
Accepted: 3 January 2022 and four or more drinks (female) in 2 h or less (NIAAA National Advisory Council 2004).
Published: 8 January 2022 Unintended consequences of binge drinking among American college students include
Publisher’s Note: MDPI stays neutral motor-vehicle crashes, campus violence, sexual assaults and date rape, psychological prob-
with regard to jurisdictional claims in lems, as well as academic difficulties such as missing and failing classes and overall lower
published maps and institutional affil- grades (Blanco et al. 2008; Hingson et al. 2009; Wechsler et al. 2000; Ehlers et al. 2018; Miller
iations. et al. 2017; Gilmore et al. 2015; Ehlke et al. 2021; An et al. 2017).
The present study explored the influence of the college drinking culture, alongside
other behavioral and social influence factors on students’ alcohol consumption behavior.
Specifically, this study implemented a naturalistic experiment with a college freshman
Copyright: © 2022 by the authors. population via a baseline evaluation at the beginning of the autumn semester, an alcohol-
Licensee MDPI, Basel, Switzerland.
prevention program as intervention and a post-experiment evaluation at the end of the
This article is an open access article
semester. To examine the role that alcohol-use cognition, attitude, expectancy and culture
distributed under the terms and
play on impacting student ability to control their drinking behavior, the study’s conceptual
conditions of the Creative Commons
framework is guided by the constructs of theory of planned behavior, outcome expectancies,
Attribution (CC BY) license (https://
social norms and college effect (or the drinking culture in a developmental phase).
creativecommons.org/licenses/by/
4.0/).
Soc. Sci. 2022, 11, 18. https://doi.org/10.3390/socsci11010018 https://www.mdpi.com/journal/socsciSoc. Sci. 2022, 11, 18 2 of 17
1.1. College Effect
Many college students experience developmentally limited alcoholism, which is char-
acterized by the transition from embracing their peers—and abandoning their parents—as
the major agents of influence (Ichiyama and Kruse 1998; Zucker 1994). Research suggests
that alcohol dependence during this age-related transitional period could be considered a
“short-duration” or “chronic and episodic” phenomenon (Vergés et al. 2012). This type of
life-stage specific alcoholism has been found to coincide with the “college effect,” which is
marked by favorable beliefs, motivations and attitudes toward experiencing their “rite of
passage” with drinking and partying among college freshmen (Lin and Carlson 2009).
The concept of “college effect” (Lin and Carlson 2009) can hence be conceptualized
as a form of social learning, where new students model their more senior counterparts’
behaviors such as excessive drinking and partying to maximize enjoyment of their college
experience. Specifically, the process of new students identifying themselves with peers
that favor heavy episodic drinking behavior is consistent with Bandura’s (1969) concept
of reciprocal determinism, a dimension of social learning theory, which suggests that a
reciprocal process of influence will take place between individuals and their environment.
Arnett (2000) introduced the term “emerging adulthood” to conceptually define the
period of transition between adolescence and adult roles. Emerging adults, especially those
who have just left high school and entered college life, are highly susceptible to the college
effect described above, as they often view drinking as an important part of their college
experience (NIAAA 2021). The college effect, as conceptualized here, could hence pose
a direct challenge to the student health objective of reducing heavy alcohol consumption
and binge drinking on college campuses. As little empirical research has examined the
college effect concept, the following research hypotheses are proposed to validate this
phenomenon in the current study context.
Hypothesis 1a (H1a). The college effect will be a positive predictor of drinking frequency at
pre-intervention phase.
Hypothesis 1b (H1b). The college effect will be a positive predictor of drinking frequency at
post-intervention phase.
Within this “college effect” framework, the discussion below presents the relevant
cognitive and affective constructs developed based on the theory of planned behavior
(Ajzen 1991, 2005). The theory of planned behavior (TPB) postulates that the intention
to perform a behavior is determined by attitudes towards the behavior, social norms
concerning the performance of the behavior and perceived behavioral control (Ajzen 1991,
2005).
1.2. Social Drinking Attitudes
Ajzen (1991) conceptualizes attitudes as the degree to which one evaluates a behavior
as positive or negative. Specifically, if an individual believes that the outcome of performing
a specific behavior is positive, then the individual would hold a positive attitude toward
the behavior and vice versa (Ajzen 1991, 2005; Montano and Kasprzyk 2015). Ajzen (1991)
also considers that an individual’s beliefs about a behavior are formed by associating the
behavior with certain characteristics, attributes, events or other behaviors.
Past studies have found that positive attitudes toward alcohol consumption were
significant predictors of alcohol use among college students (Bravo et al. 2017; Huchting
et al. 2008; Cooke et al. 2016; Trafimow 1996; DiBello et al. 2018; Dormal et al. 2018). Other
researchers have also reported that college students’ attitudes toward binge drinking were
a significant positive predictor of binge drinking intentions (Elliott and Ainsworth 2012;
Johnston and White 2003; Norman et al. 2007; Norman 2011; DiBello et al. 2018; Dormal
et al. 2018; Willis et al. 2020).Soc. Sci. 2022, 11, 18 3 of 17
Additional studies have likewise suggested that students’ positive attitudes toward
engaging in binge drinking were the strongest predictors of binge drinking intentions when
compared to the other variables of the TPB paradigm (e.g., subjective norms and perceived
behavioral control) (Norman et al. 2007; Norman 2011; DiBello et al. 2018; Willis et al. 2020).
Similarly, favorable attitudes toward alcohol use were significant and positive predictors of
drinking among college students affiliated with a college fraternity organization (Huchting
et al. 2008). To validate the theoretical link between attitude toward drinking and drinking
behavior in the current study context, the following hypotheses are tested.
Hypothesis 2a (H2a). Attitudes toward alcohol use will be a positive predictor of drinking
frequency at pre-intervention phase.
Hypothesis 2b (H2b). Attitudes toward alcohol use will be a positive predictor of drinking
frequency at post-intervention phase.
1.3. Social Drinking Norms
Prior studies that focused on alcohol consumption among college students indicated a
positive correlation between social norms regarding alcohol consumptions and intentions to
use alcohol (Carcioppolo and Jensen 2012; Halim et al. 2012; DiBello et al. 2018; DiGuiseppi
et al. 2018a, 2018b; DiGuiseppi et al. 2020). Previous studies have also demonstrated that
college students tend to over-represent alcohol consumption among their peers (Borsari
and Carey 2001; Carcioppolo and Jensen 2012; Halim et al. 2012; LaBrie et al. 2010; Kenney
et al. 2017; Cox et al. 2019). The construct of social norms in the TPB framework is
operationalized as subjective norms to reflect an individual’s beliefs about whether referent
important others would approve or disapprove the individual’s decision of engaging in a
specific behavior (Ajzen 1991).
In a meta-analysis evaluating the efficacy of the TPB, Armitage and Conner (2001)
reported that subjective norms were the weakest predictor of behavioral intentions. An-
other meta-analytic review similarly suggested that subjective norms explained little to
no variance in behavioral intentions (Hagger et al. 2002). DiBello et al. (2018) and Willis
et al. (2020) tested the predictive strength of injunctive norms (considered analogous to
subjective norms) and descriptive norms; their findings revealed that only descriptive
norms significantly predicted alcohol use among college students. Cialdini et al. (1990) con-
ceptualized descriptive social norms as an individual’s perception regarding the behaviors
of others; they asserted that descriptive norms reflect the type of social norms developed
through observation of public behaviors that are largely endorsed by a referent group.
Rivis and Sheeran’s (2003) meta-analysis study validated the hypothesis that the
relationship between descriptive norm and intention was positive and stronger among
younger samples than samples of other age groups. This hypothesis was derived from
life-span developmental psychology, which suggests that adolescents and young adults
are more pressured to conform to real or perceived social norms to gain group acceptance
(Mullen and Suls 1982; Pasupathi 1999; Castro et al. 1987; Van der Pligt and Eiser 1984).
Previous research has also indicated that perceived alcohol use of proximal others (friends)
and distal others (a typical student) had a significant impact on college students’ drinking
habits, such that students who perceived high drinking prevalence among their friends
and student body in general were more likely to drink alcohol (Halim et al. 2012; Paschall
et al. 2014; Kenney et al. 2017).
For example, past research has reported a high correlation (r = 0.70) between descrip-
tive drinking norms and students’ drinking behavior, when “friends” were considered a
normative reference group, (Carey et al. 2006). Comparatively, the correlation (r = 0.40)
was relatively lower between descriptive drinking norms and drinking behavior, when
the “typical university student” was used as a normative reference group (Cho 2006).
These findings were consistent with a meta-analysis study, which showed that students’
perceptions of their own drinking frequency were closer to that of their close rather thanSoc. Sci. 2022, 11, 18 4 of 17
distal others (Borsari and Carey 2003). Additional research also confirmed that students’
perceptions of proximal others’ drinking behavior were more similar to their own drinking
behavior, relative to their perceptions of the drinking behavior of “same-age” students
(McAlaney and McMahon 2007).
Extant research that specifically focuses on assessing the independent contribution
of proximal vs. distal descriptive norms on students’ drinking behavior remains limited.
To further validate whether perceptions of observed proximal vs. distal drinking norms
influence students’ drinking behavior, the following hypotheses are postulated.
Hypothesis 3a (H3a). Perceived descriptive proximal drinking norm will be a positive predictor of
drinking frequency at pre-intervention phase.
Hypothesis 3b (H3b). Perceived descriptive proximal drinking norm will be a positive predictor
of drinking frequency at post-intervention phase.
Hypothesis 4a (H4a). Perceived descriptive distal drinking norm will be a positive predictor of
drinking frequency at pre-intervention phase.
Hypothesis 4b (H4b). Perceived descriptive distal drinking norm will be a positive predictor of
drinking frequency at post-intervention phase.
1.4. Controlled Drinking Efficacy
An individual’s success in performing a behavior could depend on how much the
behavior is under his or her volitional control (Montano and Kasprzyk 2015). The TPB
conceptualizes perceived behavioral control as an individual’s perceived ease or difficulty of
performing a behavior (Ajzen 2005). The concept of perceived behavioral control is different
from Bandura’s (1995) construct of perceived self-efficacy, as the former emphasizes the
perceptions of control over a behavior and the latter focuses on the belief in one’s confidence
to perform a behavior (Cooke et al. 2016; Norman et al. 2007; Armitage and Conner 2001;
Norman and Hoyle 2004).
In the context of alcohol research among college students, perceived behavioral control
was conceptualized as students’ perceptions of control regarding drinking and included
both internal control factors (e.g., skills) and external control factors (e.g., environmental
pressures) (Huchting et al. 2008; Cooke et al. 2016; Norman et al. 2007; Gabbiadini et al.
2017). Meta-analysis studies suggested that perceived behavioral control uniquely con-
tributed to the prediction of both intentions and behavior (Cooke et al. 2016; Armitage
and Conner 2001). For example, Willis et al. (2020) and Gabbiadini et al. (2017) found
a significant negative relationship between perceived behavioral control and drinking
intentions, such that students who reported lower perceived behavioral control were more
likely to express intentions to consume alcohol.
Due to variations in operationalization, the direction of the relationship between per-
ceived behavioral control and alcohol consumption also varies. A meta-analysis conducted
by Cooke et al. (2016) demonstrated that these contradictory findings may be a result of
student reaction to the “perceived behavioral control” measurement. Specifically, students
may believe that alcohol consumption is a volitional action, as they underestimate the
impact of external factors (i.e., peer pressure) on their drinking behavior. To investigate
whether students’ beliefs about their ability to practice safe drinking skills influence their
drinking behavior, the current study incorporated both internal and external factors to
measure the proposed “controlled-drinking efficacy” construct. The following hypothesis
are posited to test the role of controlled-drinking efficacy in students’ drinking behavior.
Hypothesis 5a (H5a). Perceived controlled-drinking efficacy will be a positive predictor of drinking
frequency at pre-intervention phase.Soc. Sci. 2022, 11, 18 5 of 17
Hypothesis 5b (H5b). Perceived controlled-drinking efficacy will be a positive predictor of drinking
frequency at post-intervention phase.
1.5. Social Drinking Outcome Expectancies
Researchers suggested that a key factor in theories of health-related behaviors is peo-
ple’s cognitive motivation to engage in risky behaviors (Engels et al. 2005; Goldman et al.
1999). According to the alcohol expectancy theory, individuals’ experiences (vicarious and
direct) with alcohol consumption can influence their drinking-related outcome expectancies
(Goldman et al. 1999). Researchers have explored the effects of positive alcohol expectancies
such as mood enhancement, stress reduction, feelings of happiness, increased affective
expression and social facilitation—as well as negative alcohol expectancies such as aggres-
sion, sadness and depression, physical illness, poor school performance and risk of motor
vehicle crashes—on students’ alcohol consumption (Rimal and Real 2005; Goldman et al.
1999; Leigh and Stacy 1993; Cashin et al. 1998; Ham 2009; Ham et al. 2016; Papachristou
et al. 2018; Booth and Hasking 2009; Baines et al. 2016).
College students who reported higher levels of positive alcohol expectancies were
more motivated to indicate higher levels of alcohol consumption (Leigh and Stacy 1993;
Chassin and DeLucia 1996; Dunne and Katz 2015; LaBrie et al. 2008; Papachristou et al.
2018; Booth and Hasking 2009; Ham 2009; Ham et al. 2016; Baines et al. 2016). Similarly,
studies have found a negative relationship between negative alcohol expectancies and
drinking behaviors (Goldman et al. 1999; Goldberg et al. 2002; Goldman 1994; Papachristou
et al. 2018; Ham et al. 2016). Longitudinal studies also confirmed positive and negative
alcohol expectancies as predictors of drinking behaviors and drinking levels among young
adults (Goldman et al. 1999; Christiansen et al. 1989).
In a recent meta-analysis of alcohol interventions targeting college fraternity–society
students in the U.S., Scott-Sheldon et al. (2016) found that challenging students’ alcohol
expectancies could lead to reduction in alcohol consumption on episodic drinking occasions,
such as parties or weekend drinking. As more research is still needed to explore how
perceived positive drinking outcomes associated with social-interaction expectancy could
influence students’ drinking behavior, the current study chose to examine the connection
between these two variables before and after intervention via the following hypotheses.
Hypothesis 6a (H6a). Perceived positive social-drinking outcomes will be a positive predictor of
drinking frequency at pre-intervention phase.
Hypothesis 6b (H6b). Perceived positive social-drinking outcomes will be a positive predictor of
drinking frequency at post-intervention phase.
Lastly, to compare the pre-intervention and post-intervention evaluation results across
all measures, a research question is posed below:
RQ1a–f: Will pre-intervention and post-intervention scores differ with respect to (a) drink-
ing attitudes, (b) perceived proximal drinking norms, (c) perceived distal drinking norms (d)
controlled drinking efficacy, (e) drinking outcome expectancies and (f) drinking frequency?
2. Materials and Methods
This study was conducted at a large northeastern university in the U.S, where high-
risk drinking behavior has been an ongoing phenomenon for a number of years. The
study sample was represented by newly arrived students who enrolled in a multisection
“First Year Experience” course, with the cooperation and approval of the offices of the vice
provost for undergraduate studies and vice president of student affairs. The objectives of
the course are to help facilitate student transition to college life and to build a foundation
for developing a successful college career. Prior to sample recruitment, the study protocol
was approved by the Institutional Review Board that evaluates protection for human
subjects in research. A naturalistic experiment was conducted with these students in aSoc. Sci. 2022, 11, 18 6 of 17
classroom setting, in addition to a baseline evaluation at the beginning of the semester and
a post-experiment evaluation near the end of the semester. These evaluation surveys were
conducted online via a hyperlink embedded in the course website.
This naturalistic experiment incorporated the research stimulus as part of the course
topic that addressed alcohol in college life. The stimulus consisted of introducing what con-
stitutes a standard drink for different types of alcoholic beverages and class presentations of
a three-part binge-drinking prevention program. These programs include (1) Make A Drink:
having the students pour the “pretend” 80-proof alcohol that they would typically consume
into a solo cup and then have them pour the liquid into shot glasses to demonstrate how
many shots are actually in that solo cup; (2) What Happens: showing the physiological
effects (e.g., brain, heart and other health conditions), physical outcomes (e.g., physical
fights, injuries and unprotected sex) and legal ramifications (e.g., university sanctions and
local/state laws) of unsafe drinking behavior; (3) Down More Drinks: having students
engage in a role-playing game to practice controlled-drinking skills at a social event, when
pressured by other students to participate in unsafe drinking behavior.
The total number of valid survey responses was 612 for the baseline evaluation and
465 for post-experiment evaluation. Using a systematic data cleaning procedure, only those
matched cases without any data entry errors were retained. These error-free cases (N = 231),
containing the matched responses between the baseline and post-experiment evaluations,
were utilized for data analysis. The 24% attrition rate for the study (which tested more than
30 measurement items) is below the critical threshold of 40% random attrition rate deemed
necessary to maintain data validity for a college student sample in a repeated measure
study (Pan and Zhan 2020).
2.1. Definitions
The survey instrument contains the following measurement items that operationally
defined a set of theoretical concepts explicated above and were used to test the proposed
research hypotheses and questions. All scale items were measured on a 7-point scale (e.g.,
“7” = strongly agree and “1” = strongly disagree), unless otherwise specified.
2.1.1. Drinking Attitudes
Five items were used to assess the degree to which one evaluates drinking behavior as
positive or negative. Sample items included: (1) getting sick from drinking is the necessary
price to pay for having fun; (2) you won’t get alcohol poisoning, even if you do get drunk.
Responses to the five items were averaged to form the drinking attitudes variable, with
larger values indicating more positive attitudes towards episodic social drinking (pretest,
α = 0.75; post-test, α = 0.82).
2.1.2. Drinking Norms
Two separate items were utilized to assess perceived distal and proximal social-
drinking norms. For distal norm, participants were asked to estimate the percentage
of students at their university who: (1) drink at least once a month and (2) drink at least
once a week. For proximal norm, they were also prompted to offer the percentage of their
friends who: (1) drink at least once a month and (2) drink at least once a week. For each
item, participants reported a percentage between 0 and 100 (including 0, 1–9, 10–19% and
so on). Responses to these two sets of items were each averaged to create separate measures
of distal and proximal drinking norms.
2.1.3. Controlled Drinking Efficacy
Twelve items were developed to measure students’ perceived ability to control their
drinking behavior at a social drinking occasion. These items include: (1) control oneself to
reduce alcohol consumption: (2) determine the proper amount of alcohol to consume safely;
(3) only have medically safe drinks; (4) watch self to not over drink; (5) have nonalcoholic
beverages to balance the alcohol intake: (6) monitor physical reactions to adjust alcoholSoc. Sci. 2022, 11, 18 7 of 17
intake; (7) resolve not to drink to get drunk; (8) stretch the amount of time downing each
drink; (9) control the number of drinks for a specific time duration; (10) monitor one’s
mood to control alcohol intake; (11) say no to your friends if they pressure you to keep
drinking; (12) stop drinking alcohol even when your friends are still drinking. Responses
to the twelve items were averaged to form the controlled-drinking efficacy variable, with
larger values indicating a greater efficacy to practice safe-drinking skills (pretest, α = 0.98;
post-test, α = 0.99).
2.1.4. Drinking Outcome Expectancies
Individuals’ beliefs about the anticipated consequences of drinking alcohol were
measured by seven items. Participants were asked to indicate the degree to which they
believed various social-interaction outcomes would occur in a social drinking context.
Sample items included “drinking alcohol can make you feel more physically attractive”
and “drinking alcohol can make you overcome your shyness.” Responses to the seven
items were averaged to form a measure of social-drinking outcome expectancies, with
larger values indicating more desirable expectations regarding episodic social drinking
outcomes (pretest, α = 0.83; post-test, α = 0.90).
2.1.5. College Effect
At baseline, participants were asked to indicate whether a set of activities reflects
their college culture evolving around alcohol use. Six measurement items were developed
based on the conceptualization proposed by Lin and Carlson (2009). These items included:
drinking alcohol anytime they want; drinking as much alcohol as they want; going to
parties to drink alcohol, playing drinking games, partying as much as they can; getting
drunk with their friends as a rite of passage. Summed scores for all items were averaged to
construct a composite variable (α = 0.95).
2.1.6. Drinking Frequency
Participants were asked to report the number of occasions they had a drink of alcohol
in the past month. The response categories ranged from 0 occasions to 40 or more occasions
(measured via 7 incremental levels), with a higher value indicating a greater drinking
frequency.
3. Results
Table 1 presents the Pearson product–moment correlations across all study variables. It
also displays the means and standard deviations for all the paired pre- and post-intervention
variables.
Table 1. Zero-order correlations, means and standard deviations.
1 2 3 4 5 6 7 8 9 10 11 12 13
1 College effect –
2 Attitudes (Pre) 0.66 **
3 Attitudes (Post) 0.47 ** 0.63 **
4 Distal norms (Pre) 0.19 ** 0.18 ** 0.14 *
5 Distal norms (Post) 0.08 0.12 0.08 0.54 **
6 Proximal norms (Pre) 0.49 ** 0.41 ** 0.29 ** 0.35 ** 0.19 **
7 Proximal norms (Post) 0.38 ** 0.37 ** 0.30 ** 0.21 ** 0.40 ** 0.72 **
8 Control Efficacy (Pre) −0.07 −0.13 −0.08 −0.05 0.01 −0.07 −0.07
9 Control Efficacy (Post) −0.02 −0.06 −0.12 0.1 0.14 * 0.03 0.01 0.43 **
10 Expectations (Pre) 0.57 ** 0.32 ** 0.21 ** 0.14 * 0.06 0.42 ** 0.35 ** −0.12 −0.04
11 Expectations (Post) 0.42 ** 0.25 ** 0.30 ** 0.05 0.13 0.29 ** 0.36 ** −0.09 −0.1 0.45 **
12 Drinking Freq (Pre) 0.62 ** 0.50 ** 0.42 ** 0.13 * 0.09 0.57 ** 0.50 ** −0.01 0.02 0.58 ** 0.45 **
13 Drinking Freq (Post) 0.51 ** 0.41 ** 0.47 ** 0.1 0.13 0.50 ** 0.52 ** −0.03 −0.03 0.55 0.61 ** 0.76 ** –
M 2.29 2.2 2.57 69.57 63.44 60.21 55.52 5.48 5.35 1.24 1.27 3.19 2.76
SD 1.33 1.15 1.33 15.75 18.14 29.67 29.48 1.66 1.7 0.53 0.68 3.73 3.25
* p < 0.05, ** p < 0.01, two-tailed.Soc. Sci. 2022, 11, 18 8 of 17
3.1. Pre- and Post-Intervention Models
3.1.1. Pre-Intervention Model
Multiple linear regression was used to test the pre-intervention and post-intervention
models (see Table 2). Attitudes, norms, efficacy and expectancies were entered as predictors
of drinking frequency in the first step. The college effect variable was entered in the
second step to examine whether it accounted for variance above and beyond the first set of
predictors. Standardized regression coefficients are reported with alpha set to 0.05. The
pre-intervention model was significant at the first step, F (5, 204) = 32.43, p < 0.001 (R2 = 0.44;
adjusted R2 = 0.43). As hypothesized, attitudes (ß = 0.31, p < 0.001, Cohen’s f 2 = 0.07) and
proximal drinking norm (ß = 0.46, p < 0.001, Cohen’s f 2 = 0.19) were positive predictors of
drinking frequency, these findings thus confirmed H2a and H3a, respectively. However,
perceived distal drinking norm (ß = −0.08, p > 0.10, Cohen’s f 2 = 0.01), controlled drinking
efficacy (ß = 0.08, p > 0.10, Cohen’s f 2 = 0.01) and outcome expectancies (ß = 0.08, p > 0.10,
Cohen’s f 2 = 0.00) were irrelevant to drinking frequency. These results rendered H4a, H5a
and H6a without support.
Table 2. Final pre- and post-intervention regression models predicting drinking frequency.
B SE ß t VIF R2 Adjusted R2 F
Pre-Intervention Model 0.51 0.50 35.24 ***
Attitudes 0.50 0.22 0.15 2.27 ** 1.88
Norms (Distal) −0.02 0.01 −0.07 −1.35 1.15
Norms (Proximal) 0.05 0.01 0.37 6.26 *** 1.47
Efficacy 0.15 0.12 0.07 1.33 1.04
Expectancies −0.05 0.17 −0.02 −0.29 1.63
College Effect 1.05 0.20 0.37 5.28 *** 2.07
Post-Intervention Model 0.43 0.41 23.80 ***
Attitudes 0.51 0.17 0.22 2.99 ** 1.74
Norms (Distal) −0.01 0.01 −0.04 −0.69 1.26
Norms (Proximal) 0.04 0.01 0.35 5.33 *** 1.45
Efficacy 0.01 0.10 0.01 0.13 1.05
Expectancies 0.16 0.15 0.07 1.06 1.63
College Effect 0.61 0.16 0.25 3.83 *** 1.41
* p < 0.05, ** p < 0.01, *** p < 0.001.
At step 2, the addition of the college effect variable yielded a significant mode as well,
F (6, 204) = 35.24, p < 0.001 (R2 = 0.51; adjusted R2 = 0.50), accounting for an additional
6.7% of the variance (p < 0.001) and with an identical set of significant and nonsignificant
predictors relative to step 1. Specifically, while attitudes (ß = 0.15, p < 0.05, Cohen’s f 2 = 0.01)
and proximal drinking norm (ß = 0.37, p < 0.001, Cohen’s f 2 = 0.10) were positive predictors
of drinking frequency, the same is not true for perceived distal drinking norm (ß = −0.07,
p < 0.10, Cohen’s f 2 = 0.00), controlled drinking efficacy (ß = 0.07, p > 0.10, Cohen’s f 2 = 0.00)
and outcome expectancies (ß = −0.02, p > 0.10, Cohen’s f 2 = 0.00). With the college effect
variable emerging as a significant predictor of social drinking frequency (ß = 0.37, p < 0.001,
Cohen’s f2 = 0.07) at step 2, this particular result validated H1a.
3.1.2. Post-Intervention Model
Regression analysis revealed that the post-intervention model was also significant
at the first step, F (5, 189) = 23.89, p < 0.001 (R2 = 0.39; adjusted R2 = 0.37). The pattern
of findings in this initial post-intervention analysis was identical to the findings obtained
for the pre-intervention analysis. As hypothesized, attitudes (ß = 0.29, p < 0.001, Cohen’s
f 2 = 0.06) and proximal drinking norm (ß = 0.41, p < 0.001, Cohen’s f 2 = 0.14) were positive
predictors of drinking frequency, providing statistical support for H2b and H3b. Perceived
distal drinking norm (ß = −0.05, p > 0.10, Cohen’s f 2 = 0.00), controlled drinking efficacy
(ß = 0.01, p > 0.10, Cohen’s f 2 = 0.00) and outcome expectancies (ß = 0.12, p > 0.10, Cohen’sSoc. Sci. 2022, 11, 18 9 of 17
f 2 = 0.01) were not associated with drinking frequency. Thus, results failed to confirm H4b,
H5b and H6b.
After introducing the college effects variable at the second step, the final post-intervention
model was also statistically significant, F (6, 188) = 23.80, p < 0.001 (R2 = 0.43; adjusted
R2 = 0.41), accounting for an additional 4.4% of the variance (p < 0.001). Attitudes (ß = 0.22,
p < 0.01, Cohen’s f 2 = 0.03) and proximal drinking norm (ß = 0.35, p < 0.001, Cohen’s
f 2 = 0.09) were again the positive predictors of drinking frequency. The opposite is true for
perceived distal drinking norm (ß = −0.04, p < 0.10, Cohen’s f 2 = 0.00), controlled drinking
efficacy (ß = 0.01, p > 0.10, Cohen’s f 2 = 0.00) and outcome expectancies (ß = 0.07, p > 0.10,
Cohen’s f 2 = 0.00). As the college effect variable was found to be a significant predictor of
drinking frequency (ß = 0.25, p < 0.001, Cohen’s f 2 = 0.05) at step 2, this result provided
support for H1b.
3.2. Pre- and Post-Intervention Change
We conducted a series of paired samples t-tests to examine RQ1a–1f, which address
whether pre-intervention and post-intervention scores differed with respect to attitudes,
proximal norms, distal norms, efficacy, expectancies and drinking frequency. Given that
there were six tests in total, the alpha threshold was reduced to 0.01 (guided by the Bon-
ferroni correction) to avoid overly liberal rejection of multiple null hypotheses. Results
indicated a significant pre–post difference for four of the six measures. Beginning with
our criterion variable, post-intervention social drinking frequency was significantly lower
than pre-intervention social drinking frequency, t(215) = 2.49, p = 0.01, Cohen’s d = 0.17,
indicating the overall success of the intervention. With regards to the proposed determi-
nants of social drinking behavior, post-intervention attitudes were significantly higher
than pre-intervention attitudes, t(207) = 4.87, p < 0.001, Cohen’s d = −0.34. The post-
intervention proximal-drinking norm was significantly lower than the pre-intervention
norm, t(221) = 3.16, p < 0.01, Cohen’s d = 0.21. Likewise, the post-intervention perceived
distal drinking norm was significantly lower than the pre-intervention norm, t(225) = 5.62,
p < 0.001, Cohen’s d = 0.37. Although post-intervention drinking-outcome expectancies
were higher than pre-intervention drinking-outcome expectancies, the test did not reach
statistical significance when applying the reduced alpha threshold, t(209) = −2.47, p < 0.05,
Cohen’s d = −0.17. No change was observed when considering controlled-drinking efficacy,
as post-intervention scores did not differ from pre-intervention scores, t(204) = 1.05, p > 10,
Cohen’s d = 0.07. See Table 1 for means and standard deviations.
3.3. Additional Post Hoc Exploratory Analyses
Carcioppolo and Jensen (2012) differentiated between students’ perceptions of a typical
student’s drinking behavior (i.e., descriptive norms) and students’ perceptions regarding
drinking traditions on college campuses (i.e., historical drinking norms). Their study
found that both types of descriptive norms predicted drinking behavior, and historical
drinking norms also moderated the relationship between descriptive norms and drinking
behaviors. In the current study context, the college effect could be seen as an extension
of the “historical drinking norms,” as it reflects the drinking culture that the new college
students will learn and experience. To further explore and elucidate the role of the college
effect on students’ drinking behavior, a moderation analysis was conducted to ascertain
whether the college effect moderated the relations between drinking frequency and its
predictor variables.
First, we tested the baseline data, which showed that the interaction between the
college effect and proximal norms was statistically significant (ß = 0.47, p < 0.05, Cohen’s
f 2 = 0.04). The full model accounted for 58% of the variance in drinking behavior, with
approximately 1% of the variance attributed to the addition of the interaction term (p < 0.05).
By comparing participants who scored low versus high in the college effect (e.g., one
standard deviation below and above the mean), results showed that when the college effect
was low, the relationship between proximal norms and drinking behavior was positiveSoc. Sci. 2022, 11, 18 10 of 17
but weak (p < 0.001). By comparison, the same positive relationship was much stronger
when the college effect was high (p < 0.001). This indicates that proximal drinking norms
were associated with higher drinking frequency, especially for participants who scored
high on the college effect. None of the other variables, including distal norms, drinking
attitudes, controlled-drinking efficacy and drinking outcome expectancies in the model,
had a significant interaction with the college effect to predict drinking (p > 0.10).
Next, we conducted a similar analysis on the follow-up data. An identical pattern
of results was observed. The interaction between the college effect and proximal norms
was statistically significant (ß = 0.47, p < 0.01, Cohen’s f 2 = 0.03). The full model accounted
for 55% of the variance in drinking behavior, with approximately 1% attributed to the
interaction (p < 0.01). Once again, the positive relationship between proximal norms
and drinking behavior was weak (p < 0.05) when the college effect was low but gained
in strength when the college effect was high (p < 0.001). None of the other variables in
the follow-up model significantly interacted with the college effect to predict drinking
frequency.
4. Discussion
This study is among the first to implement an alcohol prevention program that directly
engages and trains students in safe-drinking skills via a naturalistic experiment in a formal
classroom setting. To assess the outcomes of this intervention, which was tailored to target
the first-year college students, a pre-test and post-test evaluation was conducted. Evaluation
results suggested that the intervention was successful in reducing social drinking behavior.
Specifically, drinking frequency (RQ1f), proximal drinking norm(RQ1b) and distal drinking
norm (RQ1c) all showed a decrease from pre-test to post-test. However, social drinking
attitudes (RQ1a) increased in their magnitude from pre- to post-intervention. No significant
change from pre-test to post-test was found for controlled drinking efficacy (RQ1d) and
drinking outcome expectancies (RQ1e).
One possible explanation for this increase in drinking attitudes is that the intervention
might have created an unanticipated effect. That is, the intervention in fact provided the
students a positive vantage point for assessing their drinking behavior from a safe-drinking
perspective. This unanticipated effect, however, seems to have been counterbalanced by
the reduced drinking frequency resulting from the intervention. It is also likely that the
lack of change in positive drinking-outcome expectancies could also have helped diffuse
the potential effects associated with the increase in drinking attitude. In other words,
even though participants viewed alcohol use more favorably at post-test, such positive
attitudes were not complemented by increased drinking frequency or expectations of
achieving desirable social-interaction consequences (e.g., overcoming shyness and feeling
more socially confident).
The success of the current intervention on decreasing students’ drinking frequency and
perceived proximal drinking norm indicates a preliminary validation of the techniques used
to engage and train the students in the experiment. Specifically, this intervention includes
having the first-year college students play the “make a drink” game to become aware of
the amount of 80-proof alcohol (i.e., the number of “shots”) they typically consume, when
a solo cup is used as the vessel for serving alcoholic beverages. Results of this exercise
almost always reveal that students severely underestimate the actual amount of alcohol
they usually consume at a social event. As a practice, this intervention module is both fun
and useful to undergraduate students.
A second intervention component involves showing students “what happens” in
an experiential mock exercise that demonstrates how heavy drinking can cause physical,
psychological and/or financial injuries, such as assaults, fights, accidents, sickness, alcohol
poisoning, arrests and legal charges, among others (Blanco et al. 2008; Hingson et al. 2009;
Wechsler et al. 2000). Even though all first-year college students received online alcohol
prevention training via the AlcoholEdu course, it is not surprising to find that students lackSoc. Sci. 2022, 11, 18 11 of 17
a good understanding of the real negative consequences of unsafe drinking behaviors until
they have personally experienced them.
The third component of the intervention allows students to roleplay different types of
safe-drinking skills, including sipping, eating food before drinking and between drinks,
taking breaks between sips, adding a soft drink to the alcoholic beverage, safeguarding the
drink, staying with friends and alcohol refusal techniques, among others. Again, it is not
uncommon to find that students are utterly unaware of these “common-sense” safety tips
that are not difficult to acquire and can lead to good safe-drinking habits.
Turing to the pre-test and post-test regression models that tested the predictors of
drinking frequency, both models generated an identical pattern of results entailing the
same set of significant and nonsignificant predictors. Specifically, more positive social
drinking attitudes (H2a–b) were a significant predictor of greater drinking frequency last
month, alongside college effect (H1a–b) and proximal norm (H3a–b). However, the same
is not true for distal norms (H4a–b), controlled-drinking efficacy (H5a–b) and outcome
expectancies (H6a–b), as they were not found to be significant predictors of drinking
frequency. With regard to the post hoc moderator analysis, the college effect emerged as a
positive moderator for the influence of proximal norms on drinking frequency. The college
effect did not significantly moderate the influence of other predictor variables on drinking
frequency, however.
As proximal norms instead of distal drinking norms were a significant predictor of
the two criterion variables, pre- and post-intervention episodic drinking behavior, these
findings confirmed the results of some past research and contrasted that of others (Halim
et al. 2012; Campo et al. 2003; Granfield 2005; Licciardone 2003; Lewis and Neighbors
2004; Perkins and Craig 2006; Thombs et al. 2007). In essence, the findings suggested
that while students’ drinking frequency was significantly influenced by their friends’
drinking behaviors, a typical peer’s drinking behaviors at school was not relevant in this
context. By implication, these findings may also provide a partial explanation for why some
social norms campaigns that focused on correcting the perceived distal descriptive norms
might have failed (Paschall et al. 2006). These contradictory findings are not necessarily
detrimental to the validity of these social norms campaigns, as variations in study settings
could also generate different study results.
Findings further indicated that social drinking outcome expectancies were irrelevant
to drinking frequency for both pre- and post-intervention evaluations. These results are
inconsistent with past research findings (e.g., Papachristou et al. 2018; Ham et al. 2016;
Baines et al. 2016). This contradiction could have been a function of the measurement items
used in the current study for assessing outcomes expectancies. Specifically, these measures
gauged student anticipation of gaining the desired social status and sociability outcomes
through consuming alcohol in a party setting. For this reason, participants might have been
in denial of their reliance on alcohol to achieve their social interaction efficacy. To put it
another way, if participants could acknowledge how they rely on heavy episodic drinking
to feel more socially adequate or integrated, then this admission may help them consider
such harmful consequences as drunken brawls, injury and unprotected sex more carefully.
Interestingly, results also showed that controlled-drinking efficacy was not a significant
predictor of pre- and post-intervention drinking frequency. These findings may reflect
an optimistic bias whereby students may believe that they are in control of their alcohol
consumption, while overlooking external influences such as social drinking norms and
the college effect. For example, according to Cooke et al. (2016), students may consider
college drinking behavior as an outcome of their own volition, which is not subject to
other environmental influences (e.g., peer pressure). Another plausible explanation could
be that students who are regular episodic drinkers may envision themselves as having
sufficient control over their alcohol consumption behavior, as they perceive negative
drinking outcomes to be normal and accepted as part of the partying scene (Lin et al. 2014).
It is also possible that as a majority of the students were engaged in episodic social drinking,Soc. Sci. 2022, 11, 18 12 of 17
the intervention may not have been compelling enough to change student belief about their
own controlled-drinking efficacy.
As for the role of the college effect played in students’ drinking behavior, the findings
clearly suggested that the college-effect related mindset, which might have reflected a
hedonistic drinking culture, was at display here (Bravo et al. 2017; Cooke et al. 2016;
Norman et al. 2007; Norman 2011). As the results of the moderation analyses singled
out the college effect as a positive moderator only between proximal norms and drinking
frequency, this suggests that when the college effect was stronger, so was the impact
of friends’ drinking norms on social drinking frequency. These findings also broadly
confirmed Carcioppolo and Jensen’s (2012) conception of historical drinking norms, which
describes a type of descriptive norms that is reflected in the drinking traditions on college
campuses. Based on the findings associated with the college effect, a construct that is
conceptualized to be related to developmental-stage attitudes toward alcohol use, it is clear
that the college effect could be an important factor for motivating episodic drinking activity
on college campuses.
It is worth noting that even though the constructs of attitudes, outcome expectancies
and efficacy associated with alcohol-use behavior (Bravo et al. 2017; Huchting et al. 2008;
Cooke et al. 2016) are not necessarily new, an innovative “reality-based” approach was
adopted here to conceptualize these measures from the students’ point of view. In particular,
this approach invites students to evaluate their drinking attitudes to include a proxy rite of
passage measure; it also enables students to express their drinking-outcome expectancies
via the desired social interaction efficacy in a social setting. As such, the attitude measure
reflects an air of invincibility and inevitability, as it relates to why one chooses to drink and
enjoys alcohol use without a care in the world. In a similar vein, the construct of drinking
outcome expectancies represents those positive social consequences that one would hope
to experience. This conceptual approach is particularly suited for an intervention study
that intends to make an impression on the students who need to know how to stay safe
if they choose to drink alcohol at a party. By the same token, the construct of drinking
efficacy is conceptualized as a “controlled” drinking efficacy. Through focusing on the
“controlled” aspect of drinking behavior, this construct again enables the intervention to
emphasize the attainable safe-drinking skills that students could learn to practice at a social
drinking event.
As the current study reconceptualized the construct of drinking attitudes, outcome
expectancies and self-control efficacy through a “safe-drinking” intervention, it also intro-
duced the college effect construct stemming from the framework of developmental-stage
alcoholism. Under this framework, an individual does not typically evaluate the negative
outcomes or implications of episodic alcohol use, social drinking norms or self-control
efficacy to manage their drinking behavior for harm avoidance in a rational manner. Hence,
the underlying theoretical assumptions linking all the variables under study give per-
mission to an alternative approach that is distinct from the “rational decision-making”
paradigm articulated by the TPB. This alternative approach instead assesses, explains and
intervenes in the unsafe drinking behavior from the standpoint of a hedonistic drinking
culture (Bravo et al. 2017; Cooke et al. 2016; Norman et al. 2007; Norman 2011), consistent
with the historical drinking norms (Carcioppolo and Jensen 2012) and the characteristics of
the college effect.
Taken together, these study findings suggest that when the first-year college students
attend a social event, they choose to consume alcohol because drinking is an important
part of college culture. These college students are also willing to accept both the good
(e.g., being social) and the bad (e.g., feeling sick from over drinking) outcomes associated
with the college effect, since attending college is a once in a lifetime opportunity. As
these college students’ alcohol consumption activities fall into a form of developmental
alcoholism, they do not necessarily internalize the “locus of control” that will help guide
them to develop meaningful forethoughts to rationally evaluate the risks and benefits
of the real-world consequences. Hence, when addressing the episodic heavy-drinkingSoc. Sci. 2022, 11, 18 13 of 17
problem among college students, it would be useful to consider the cognitive, affective
and behavioral aspects of this phenomenon from the students’ perspective. This approach
would better enable researchers and practitioners alike to develop evidence-based tailored
communication strategies to more effectively target these emerging adults with achievable
and desirable positive health outcomes.
4.1. Study Limitations
Several limitations are noted for this study, as follows. First, the current research tested
a conceptual framework with selective measures to validate the experimental results. Ad-
ditional measures related to other developmental factors such as emotional and personality
factors could be added as potential moderators to explore more advanced theoretical expla-
nations. Second, a set of measures was developed to test both existing and new theoretical
concepts that were incorporated in an innovative study framework. Even though these
measures all exhibited satisfactory measurement validity and reliability, further refinement
could potentially improve their replicability in relevant studies that are conducted with a
similar or different methodology. Third, study results were generated through a pre- and
post-measure one-shot field experimental design. To verify these findings, longitudinal
studies that replicate the current research design would be required.
4.2. Implications for Research and Best Practices
While the intervention did not generate a statistically significant result in altering
student attitudes or expectancies toward alcohol consumption, these results are not wholly
unexpected for exposure to a one-shot only intervention. Realistically speaking, changing
student attitudes toward alcohol use is not a goal that is attainable through this type of one-
shot intervention, as alcohol consumption has been both widely prevalent and celebrated
in American society and Western culture for centuries.
Importantly, the finding that indicates a decrease in student perception of other stu-
dents’ drinking behavior, including both perceived distal and proximal drinking norm,
is a highly significant outcome of the intervention. As perceived descriptive norms as-
sociated with alcohol use are both ascribed by the strength of students’ belief as well as
reflected by the prevailing episodic drinking routine on campus, a change in perceived
descriptive drinking norms may signal a potential conversion to endorse a lower social
drinking frequency in the future.
The failure of the intervention to significantly increase student belief about their ability
to practice safe-drinking skills was, again, not entirely unanticipated. Being acquainted
with safe-drinking skills does not guarantee the application of these skills when students
are participating in a party and wish to act the way that their friends and peers do. This is
especially true, as the college effect could make students feel compelled to consume alcohol
the way others do, for the purpose of demonstrating their willingness to “be social” at
parties and gaining acceptance to establish their group identity. This type of thinking and
reality then brings forth a challenge for the dominant alcohol-prevention paradigm, which
emphasizes “teaching” and “correcting” students’ misperceptions about the exaggerated
student drinking norms.
Compared to recent research that investigated the effectiveness of mandatory alcohol
prevention programs (implemented as a self-administered online alcohol education course),
the current prevention approach is either equally or more effective in reducing the cognitive,
affective and/or behavioral outcomes of episodic drinking behavior among the first-year
students. For instance, an online prevention program such as Alcohol-Wise was found to
reduce drinking frequency in a two-campus study, though additional research is needed to
validate the study findings (Croom et al. 2015). However, the most widely administered
online prevention program, AlcoholEdu, has continued to receive mixed results. For
example, a study examining the impact of the course on reducing first-year drinking
frequency found no support for its effectiveness (Paschall et al. 2014), as another study
reported contradictory results (Lovecchio et al. 2010).Soc. Sci. 2022, 11, 18 14 of 17
Future research could consider developing alternative measurement items to assess
the constructs that have been commonly used to discover potentially new and theoretically
meaningful conceptual dimensions. For instance, a further exploration of the college effect
construct may yield additional insights that are useful but currently unavailable. Empirical
research should also consider applying the developmental-stage alcoholism perspective to
study heavy episodic college drinkers to develop tailored and timely intervention, as these
heavy drinkers may not mature out of their alcohol abuse after graduation.
In addition, adopting new communication technology such as a mobile phone app
platform could be a productive approach to collect real-time and longitudinal data. For
example, Merrill et al. (2018) utilized mobile texting in a pilot experiment to examine drink-
ing behaviors, norms (injunctive and descriptive) and consequences. Study participants
rated this messaging approach, which sent multiple text messages to their phone, to be
highly acceptable.
Future studies could also consider developing a mobile phone app that would allow
students to continuously interact with a set of features/icons to report their mood state and
alcohol-use related activities. These types of data could be calibrated to prompt the mobile
app to provide a real-time warning when the student is on the verge of entering a high-risk
state emotionally and/or behaviorally. The data collected via the mobile app could also be
applied as a self-reflection tool and incorporated into a one-on-one intervention session
such as a motivational interview.
For practitioners, the one-shot intervention reported here serves as an example of
an alternative intervention approach, relative to the conventional intervention strategy
which often places its emphasis on broadly focused social norms campaigns, online training
modules (e.g., AlcoholEdu) or judicially sanctioned education programs (e.g., BASICS). In
particular, this multistage training routine, aiming at harm avoidance and harm reduction,
could be easily implemented through interacting with students who share the same living
quarters or student club membership in a group setting. Practitioners could consider
implementing similar or the same set of intervention techniques (see presented here) to
invite students to learn and practice safe-drinking skills in the first year of their college life.
Author Contributions: Conceptualization, C.A.L.; methodology, C.A.L.; validation, C.A.L., J.L.C.
and A.B.B.; formal analysis, J.L.C. and C.A.L.; writing—original draft preparation, C.A.L., J.L.C. and
A.B.B.; writing—review and editing, C.A.L., J.L.C. and A.B.B.; project administration, C.A.L. All
authors have read and agreed to the published version of the manuscript.
Funding: This research received no external funding.
Institutional Review Board Statement: The study was conducted in accordance with the Declaration
of Helsinki, and approved by the Institutional Review Board (or Ethics Committee) of University of
Connecticut (protocol code H07-155, approved 25 June 2009).
Informed Consent Statement: Informed consent was obtained from all subjects involved in the
study.
Data Availability Statement: The data are available upon request.
Conflicts of Interest: The authors declare no conflict of interest.
References
Ajzen, Icek. 1991. The theory of planned behavior. Organizational Behavior and Human Decision Processes 50: 179–211. [CrossRef]
Ajzen, Icek. 2005. Attitudes, Personality, and Behavior. London: McGraw-Hill Education (UK).
An, Brian P., Chad N. Loes, and Teniell L. Trolian. 2017. The relation between binge drinking and academic performance: Considering
the mediating effects of academic involvement. Journal of College Student Development 58: 492–508. [CrossRef]
Armitage, Christopher J., and Mark Conner. 2001. Efficacy of the theory of planned behaviour: A meta-analytic review. British Journal
of Social Psychology 40: 471–99. [CrossRef]
Arnett, Jeffrey Jensen. 2000. Emerging adulthood: A theory of development from the late teens through the twenties. American
Psychologist 55: 469. [CrossRef]
Baines, Laura, Andrew Jones, and Paul Christiansen. 2016. Hopelessness and alcohol use: The mediating role of drinking motives and
outcome expectancies. Addictive Behaviors Reports 4: 65–69. [CrossRef] [PubMed]You can also read
